Clustering of behavioral spatial trajectories in neuropsychological assessment

Neuropsychological assessment consists in the evaluation of cognitive skills based on subjects’ performance. Here we present an integrated procedure based on an enhanced version of a spatial cognition task, the Baking Tray Task (BTT). In this version, data are automatically recorded and immediately available for the analysis.We then develop a two step statistical analysis to identify groups of homogeneous patterns of performance: (1) a functional principal components analysis represents the data as points on a subspace; (2) a finite mixture model-based clustering algorithm detects groups having similar patterns of performance. We present the results from a random sample of 114 healthy subjects. Future research will aim to offer diagnostic procedures for spatial deficits, based on these methods.